Predicting the usefulness of online lecture reviews: using LDA and machine learning†
Kyungmin Kwon1 · Shinhaeng Lee2 · Myeongjin Lee3 · Hanjun Lee4
1Graduate School of Information, Yonsei University
234Department of Management Information Systems, Myongji University
234Department of Management Information Systems, Myongji University
Correspondence to: † This work was supported by 2023 Research Fund of Myongji University.
1 Graduate student, Graduate School of Information, Yonsei University, Seoul, 03722, Korea.
2 Student, Department of Management Information Systems, Myongji University, Seoul 03674, Korea.
3 Student, Department of Management Information Systems, Myongji University, Seoul 03674, Korea.
4 Associate professor, Department of Management Information Systems, Myongji University, Seoul 03674, Korea. E-mail: hjlee1609@gmail.com
1 Graduate student, Graduate School of Information, Yonsei University, Seoul, 03722, Korea.
2 Student, Department of Management Information Systems, Myongji University, Seoul 03674, Korea.
3 Student, Department of Management Information Systems, Myongji University, Seoul 03674, Korea.
4 Associate professor, Department of Management Information Systems, Myongji University, Seoul 03674, Korea. E-mail: hjlee1609@gmail.com
Received February 4, 2024; Revised March 5, 2024; Accepted March 20, 2024.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
- Abstract
- In online review sites, the problem of information overload due to too many reviews frequently occurs. Accordingly, useful reviews are recommended through reviewer voting, but there exist limitations. Therefore, this study proposes a model to predict the usefulness of reviews on lecture review sites. To this end, we collected 44,845 online lecture reviews and used Latent Dirichlet Allocation (LDA) to extract various variables, including the proportion of topics included in each review, and learn them through machine learning algorithms. Among the models we developed, the XGBoost-based model showed the best performance with an accuracy of 87.96 percent. The approach and results of this study are expected to provide meaningful implications for solving the problem of information overload in online review sites.
- Keywords : Everytime, LDA, lecture review, machine learining, review helpfulness
Full Text(PDF) Free
